[
https://issues.apache.org/jira/browse/HBASE-29291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17950076#comment-17950076
]
Shanmukha Haripriya Kota commented on HBASE-29291:
--------------------------------------------------
*Operational Considerations:*
*Missing-File Read Failures:*
Until a replica cluster has run both `refresh_hfiles` and `refresh_meta`, any
splits/compactions in the active cluster may leave its meta table pointing at
deleted HFiles or old region directories. Reads against those stale paths will
throw:
{*}FileNotFoundException{*}: File does not exist: /…/old-region/…
{*}HFileNotFoundException{*}: Failed to open StoreFile at …
To prevent these FNF read failures on read replica cluster, in the future we
can extend this refresh_meta to an automated meta table sync.
> Add a command to refresh/sync hbase:meta table
> ----------------------------------------------
>
> Key: HBASE-29291
> URL: https://issues.apache.org/jira/browse/HBASE-29291
> Project: HBase
> Issue Type: Sub-task
> Components: API, meta, shell
> Reporter: Shanmukha Haripriya Kota
> Assignee: Shanmukha Haripriya Kota
> Priority: Major
>
> The objective of this *refresh_meta* command is to synchronize the
> *_hbase:meta_* table with the actual state of regions in the backing storage.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
